| 1 |
Hi Nick, |
| 2 |
|
| 3 |
thanks for your contribution. |
| 4 |
|
| 5 |
I think there is no general answer to your question. It depends if it |
| 6 |
is a disadvantage of running the mpi-version with 1 MPI process over |
| 7 |
running the serial version. I imagine that LAMMPS enable some special |
| 8 |
optimization if mpi is disabled as it can assume that everything |
| 9 |
happens locally and no mpi message need to be send. On the other hand, |
| 10 |
if the build of LAMMPS takes hours, one might prefer building it only |
| 11 |
once. Do LAMMPS offer other palatalizations like OpenMP or Threads to |
| 12 |
use multiple cores? |
| 13 |
|
| 14 |
From a brief look at our patch I see the following issues: |
| 15 |
- I am not sold on this Makefile patch and patching of the patch |
| 16 |
approach, seems a bit fragile. It might be better to use the default |
| 17 |
gcc Makefile and patch that. I prefer adding CC,CFLAGS etc, on emake |
| 18 |
comandline, you could create a function to avoid writing it on every |
| 19 |
emake call. (my_emake() { emake "$@" CC= CFLAGS=; } ) |
| 20 |
- sed calls need to secured with "|| die" |
| 21 |
- DEPEND should be static, DEPEND="mpi? ( vitrual/mpi )" should do it |
| 22 |
- virtual/fortran is missing in DEPEND or use fortran-2.eclass and |
| 23 |
FORTRAN_NEEDED="fortran" instead. |
| 24 |
- lammps-gzip enables normal gzip support, so the use flags should be |
| 25 |
called gzip. |
| 26 |
|
| 27 |
If you are unsure about some technical details in an ebuild, create a |
| 28 |
pull request on github and somebody will give you advice in the |
| 29 |
review. |
| 30 |
|
| 31 |
Otherwise good job, looking forward to more commits from you, |
| 32 |
|
| 33 |
Christoph |
| 34 |
|
| 35 |
2013/5/12 Nicolas Bock <nicolasbock@×××××.com>: |
| 36 |
> Hi, |
| 37 |
> |
| 38 |
> I recently added a lammps ebuild to the science overlay. It so far only |
| 39 |
> builds the serial version, and has very limited support for build time |
| 40 |
> options. I now added the mpi use flag and changed the ebuild so that it |
| 41 |
> builds the mpi version. I have attached a patch against master that does |
| 42 |
> this. I would like to get some feedback on those changes. Are they |
| 43 |
> acceptable in this form? What should I have done differently? How would I go |
| 44 |
> about writing an ebuild that builds both, the serial _and_ the mpi version |
| 45 |
> in case the mpi use flag is set? |
| 46 |
> |
| 47 |
> Thanks already, |
| 48 |
> |
| 49 |
> nick |
| 50 |
> |
| 51 |
|
| 52 |
|
| 53 |
|
| 54 |
-- |
| 55 |
Christoph Junghans |
| 56 |
http://dev.gentoo.org/~ottxor/ |
Archives